Andrew J. Brown - (contact?) - ETC - amazing puter blunder creates 80+ fantasy bogus apparatus entries

Steven Avery

Steven Avery

Administrator
  • Andrew J. Brown8/16/2022 2:10 am
    Fictitious citations from codex 0184 in the “Editio Critica Maior” (ECM) of the Greek text of the Gospel of Mark, printed in 2021.

    Those who consult the printed apparatus of the “Editio Critica Maior” at Mark chapter 15 will find, among much other information, an unexpectedly long series of citations from codex 0184. In the standard lists of Greek New Testament manuscripts, 0184 is described as a sixth-century bilingual majuscule fragment (currently located in Vienna), containing part of Mark chapter 15, verses 36-37 and 40-41. Similar details of the contents of this manuscript are given near the beginning of the second ECM volume on Mark (“Supplementary Material”, page 6). On the website of the Institut für Neutestamentliche Textforschung (INTF), there is also a photograph of this fragment, accompanied by a transcription of the Greek words and letters which are legible in those four verses.

    It was therefore surprising to see, in the printed ECM apparatus of manuscript readings, more than eighty citations from codex 0184 in the earlier part of Mark 15, from verse 1 through to verse 35. These readings initially made it seem that an important discovery had been made of a new portion of 0184, and that its contents were being published in the ECM for the first time. Even more remarkable was the fact that, when these new citations from 0184 are examined in detail, they show a 100% agreement with the “A-text” (the editors’ preferred text), apart from the curious “omission” of verse 14. According to the printed ECM apparatus, codex 0184 (without the slightest deviation of spelling, or any missing words or letters, and without any hesitation over its decipherment) is thus a perfect replica of the main ECM text in Mark 15, verses 1-13 and 15-35. This was all the more astonishing, as not even the famed codex Vaticanus (one of the witnesses on which the ECM edited text chiefly depends) managed to achieve such a result. By contrast, in the previously-known portion of 0184, there are two places where this fragment disagrees with the ECM main text (in verses 36 and 40 of chapter 15), and some of its letters are wholly or partly illegible.

    Users of the printed edition should be informed that no “new” fragment of codex 0184 has actually been found. It is possible that another piece of this manuscript might one day be discovered, but at present all the references to codex 0184 in verses 1-13 and 15-35 are simply mistaken, and should be deleted. The inclusion of these erroneous citations has the effect of artificially inflating the claimed degree of support for readings adopted in the main text of the ECM. It appears that a single human error in the treatment of the manuscript’s lacunae may have led, in turn, to the creation of more than eighty further errors by the ECM computer programme, which was responsible for this whole series of false citations from codex 0184.

    Andrew J. Brown
    Reply
    Replies

    1. Reply

  • blank.gif

    Andrew J. Brown8/16/2022 2:12 am
    (Fictitious citations from codex 0184, continued)

    A potential clue to the origin of these errors is visible in the detailed list of lacunae, given on pages 17-21 of the second volume (“Supplementary Material”). It is noticeable, on page 19, that the lacunae for codex 0184 are said to include Mark chapter 10, verse 48, and chapter 15, verse 14. While the printed apparatus cites readings from 0184 in the verses which precede and follow this second “lacuna”, no readings seem to be cited from 0184 in chapter 10. If the manuscript contains no part of chapter 10, it cannot rightly be said to suffer only from a one-verse lacuna in that chapter. Leaving aside this anomaly, a reader who looks across to the next column of page 19 will observe that an identical pair of lacunae is attributed to codex 184, a thirteenth-century minuscule. There would thus seem to be a distinct possibility that someone who was momentarily confused by the similarity between the numerals 0184 and 184 might, with a few keystrokes, have caused the lacunae of the later manuscript to be incorrectly combined with the lacunae of the other manuscript that was 700 years older. Since a genuine fragment of codex 0184 contains most of verse 36 of chapter 15, any misleading information about the existence of a lacuna at verse 14 might have caused the computer programme or database to conclude (wrongly) that other verses of this chapter, from 1 to 13 and from 15 to 35, were present in the manuscript.

    In this long passage, the only divergence between codex 0184 and the main text consists of the single fact that all these verses are missing from the manuscript. Once that essential piece of information had been removed or obscured, the computer programme or database proceeded to the logical (but incorrect) conclusion that the manuscript agreed with the main edited text at every reading shown in the apparatus, since no other divergences had been recorded. This would account for the apparent 100% agreement between codex 0184 and the so-called “A-text” in the printed apparatus for these verses. The editors of the ECM will no doubt be able to confirm in due course whether this explanation is correct, or whether there was some other reason for the error.

    It would be unfair to blame a particular individual for such a mistake. However, would it not be advisable for the editors of ECM to make use of a simple computer programme which cross-checks between the list of lacunae and the list of chapters and verses which a manuscript is already known to contain? Such a programme could also cross-check with the verse numbers recorded in the textual transcriptions which have been made from each manuscript into the database. In this way, irrelevant lacunae would be detected and discarded, and the ECM would be better protected against the risk of introducing a significant quantity of false citations into the apparatus. The much-heralded “Coherence Based Genealogical Method” was probably not designed to detect this kind of error. The set of readings which was attributed to codex 0184 should have been immediately identified as being “too coherent to be true”.

    A related problem is that the “Text-Critical Commentary”, provided in the third ECM volume on Mark (“Studies”, page 31), appears to rely on these “new” readings of codex 0184 in calculating the small number of “A-related witnesses”: at Mark 15, verses 24, 29 and 34. The reliability of the statistics is undermined by the incorrect inclusion of 0184. The degree of manuscript support for those readings was already relatively weak, and is weakened still further by the removal of codex 0184 from the argument.

    Andrew J. Brown
    Reply
    Replies

    1. Reply

  • blank.gif

    Andrew J. Brown8/16/2022 2:15 am
    (Fictitious citations from codex 0184, conclusion)

    Today perhaps a computer can be pardoned for innocently inventing more than eighty readings in a non-existent manuscript fragment. Nevertheless it demonstrates once again how the use of computerised technology, despite all its benefits, is capable of producing new kinds of error that were not previously dreamed of.

    Until quite recently this series of erroneous readings was also to be found in the version of the “Editio Critica Maior” that is displayed on the INTF website. Although most of them have now been expunged from the online edition, a remnant of these errors can still (at the time of writing) be seen online at Mark 15: 35, where codex 0184 is explicitly cited for three readings, despite not being extant in that verse. Unfortunately, whatever changes are made to the online edition, the errors cannot so easily be deleted from the printed edition that has already been circulated widely to libraries and researchers.

    [Most of the above points were communicated on 13 July 2022 to the ECM editors, who hope to make the necessary corrections relating to this manuscript in a revision of the printed volume]

    Andrew J. Brown
 
You must log in or register to reply here.
Top